Oracle 数据透视转换到 SQL 服务器

Question

简短的问题，对 SQL server 不太熟悉，但我在 oracle sql server 中有多个聚合的查询。

多重聚合的sql server等效查询是什么？

Oracle 代码：

[SELECT *
   FROM (SELECT d.loc_area,
                d.loc_rack,
                d.loc_height,
                d.loc_place,
                d.locvrc,
                row_number() OVER(PARTITION BY d.loc_area, d.loc_rack, d.loc_place
                                   ORDER BY d.loc_height) rn_pivot
           FROM (select locvrc,
                        SUBSTR(stoloc, 0, 2) loc_area,
                        SUBSTR(stoloc, 4, 2) loc_rack,
                        SUBSTR(stoloc, 6, 1) loc_place,
                        SUBSTR(stoloc, -2) loc_height
                   FROM locls) d)
  PIVOT (MAX(loc_height) AS EXT, MIN(locvrc) AS UNIQ FOR rn_pivot IN ('1' AS HEIGHT_1, '2' AS HEIGHT_2, '3' AS HEIGHT_3, '4' AS HEIGHT_4))
  ORDER BY LOC_AREA,
        LOC_RACK,
        LOC_PLACE]

SQL 服务器：

SELECT *
   FROM (SELECT d.loc_area,
                d.loc_rack,
                d.loc_height,
                d.loc_place,
                d.locvrc,
                row_number() OVER(PARTITION BY d.loc_area, d.loc_rack, d.loc_place
                                   ORDER BY d.loc_height) rn_pivot
           FROM (select locvrc,
                        SUBSTRing(stoloc, 0, 2) loc_area,
                        SUBSTRing(stoloc, 4, 2) loc_rack,
                        SUBSTRing(stoloc, 6, 1) loc_place,
                        right(stoloc, 2) loc_height
                   FROM locls) d) A
  PIVOT (MAX(loc_height) FOR rn_pivot IN ([1],[2],[3],[4])) P1 

这工作正常，但我还需要另一个聚合 |(MIN(locvrc)| 也

提前谢谢你！

溴 一月

/编辑，

原表：

 Stoloc          locvrc
 5D-32A-00       699787
 5D-32A-10       439567
 5D-32A-20       429456
 5D-32A-30       122172
 5D-32B-00       328311
 5D-32B-10       247422
 5D-32B-20       133244
 5D-32B-30       464978

子字符串后的表格：

loc_area loc_rack loc_place loc_height locvrc
 5D         32       A        00       699787
 5D         32       A        10       439567
 5D         32       A        20       429456
 5D         32       A        30       122172
 5D         32       B        00       328311
 5D         32       B        10       247422
 5D         32       B        20       133244
 5D         32       B        30       464978

我对 Oracle 的看法是：

loc_area loc_rack loc_place Height1 Uniq1   Height2 Uniq2   Height3 Uniq3 Height4 Unique4
5D       32        A         00     699787    10    439567    20    429456    30   122172
5D       32        B         00     328311    10    247422    20    133244    30   464978

因此，从原始表格中，stoloc 将被切割成不同的列。在这一行之后，区域、机架、位置和它们在一行中的高度和 locvrc。我已经在 Oracle 数据库中完成了这项工作，但不知道如何在 SQl 服务器中完成。不需要这种解决方案，任何人都可以。

Answer 1

您可能会觉得这很奇怪，但据我所知，这是我们在 SQL server 中可以做到的。

我们需要通过附加任何内容为相同的数据透视列创建另一个维度，并将其放入PIVOT 子句中。

如您所见，我已添加 1 并创建另一列用于 locvrc

SELECT *
   FROM (SELECT d.loc_area,
                d.loc_rack,
                d.loc_height,
                d.loc_place,
                d.locvrc,
                row_number() OVER(PARTITION BY d.loc_area, d.loc_rack, d.loc_place
                                   ORDER BY d.loc_height) rn_pivot,
                cast(row_number() OVER(PARTITION BY d.loc_area, d.loc_rack, d.loc_place
                                   ORDER BY d.loc_height) as varchar) + '1' rn_pivot_locvrc
           FROM (select locvrc,
                        SUBSTRing(stoloc, 0, 2) loc_area,
                        SUBSTRing(stoloc, 4, 2) loc_rack,
                        SUBSTRing(stoloc, 6, 1) loc_place,
                        right(stoloc, 2) loc_height
                   FROM locls) d) A
  PIVOT (MAX(loc_height) FOR rn_pivot IN ([1],[2],[3],[4])) P1
  PIVOT (MAX(locvrc) FOR rn_pivot_locvrc IN ([11],[21],[31],[41])) P2

附：我的目的是告诉你如何PIVOT 多个列，但不幸的是我当然无法验证结果集。

编辑：- 使用 Op 的实际数据更新查询

SELECT loc_area,loc_rack,loc_place
      ,max([1]) as height_1
      ,max([11]) as unique_1
      ,max([2]) as height_2
      ,max([21]) as unique_2
      ,max([3]) as height_3
      ,max([31]) as unique_3
      ,max([4]) as height_4
      ,max([41]) as unique_4
   FROM (SELECT d.loc_area,
                d.loc_rack,
                d.loc_height,
                d.loc_place,
                d.locvrc,
                row_number() OVER(PARTITION BY d.loc_area, d.loc_rack, d.loc_place
                                   ORDER BY d.loc_height) rn_pivot,
                cast(row_number() OVER(PARTITION BY d.loc_area, d.loc_rack, d.loc_place
                                   ORDER BY d.loc_height) as varchar) + '1' rn_pivot_locvrc
           FROM (select locvrc,
                        SUBSTRing(stoloc, 0, 2) loc_area,
                        SUBSTRing(stoloc, 4, 2) loc_rack,
                        SUBSTRing(stoloc, 6, 1) loc_place,
                        right(stoloc, 2) loc_height
                   FROM locls) d) A
  PIVOT (MAX(loc_height) FOR rn_pivot IN ([1],[2],[3],[4])) P1
  PIVOT (MAX(locvrc) FOR rn_pivot_locvrc IN ([11],[21],[31],[41])) P2
group by loc_area,loc_rack,loc_place

或

使用aggregation和case也可以做到，

SELECT loc_area,loc_rack,loc_place
      ,max(case when rn = 1 then loc_height end) height_1
      ,max(case when rn = 1 then locvrc end) unique_1
      ,max(case when rn = 2 then loc_height end) height_2
      ,max(case when rn = 2 then locvrc end) unique_2
      ,max(case when rn = 3 then loc_height end) height_3
      ,max(case when rn = 3 then locvrc end) unique_3
      ,max(case when rn = 4 then loc_height end) height_4
      ,max(case when rn = 4 then locvrc end) unique_4
   FROM (SELECT d.loc_area,
                d.loc_rack,
                d.loc_height,
                d.loc_place,
                d.locvrc,
                row_number() OVER(PARTITION BY d.loc_area, d.loc_rack, d.loc_place
                                   ORDER BY d.loc_height) rn
           FROM (select locvrc,
                        SUBSTRing(stoloc, 0, 2) loc_area,
                        SUBSTRing(stoloc, 4, 2) loc_rack,
                        SUBSTRing(stoloc, 6, 1) loc_place,
                        right(stoloc, 2) loc_height
                   FROM locls) d) A
group by loc_area,loc_rack,loc_place;